Knowledge tracking method based on dual path matrix decomposition
Through dual path matrix decomposition technology, combined with clustering model, students' answering information and learning behaviors are analyzed, the problem of low prediction accuracy in the existing knowledge tracking model is solved, and more accurate analysis of knowledge mastery and learning guidance is achieved.
Patent Information
- Application Number
- CN202411191786.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-28
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2044-08-28
AI Technical Summary
When predicting students' future answering performance, existing knowledge tracking models usually only focus on the difficulty of the exercise or the learner's knowledge level, ignoring learning behavior and style factors, resulting in a decrease in the accuracy of the prediction results.
Using a method based on dual path matrix decomposition, by collecting students' answer information and learning behavior data, building exercises and learners parameter models, deeply analyzing the degree of knowledge mastery, and combining clustering models to generate learning guidance information.
It realizes more accurate attribution analysis of the degree of knowledge mastery, can more comprehensively represent the learners' learning situation, help teachers adjust teaching plans, and help students identify learning shortcomings and plan more effective learning plans.
Smart Images

Figure CN119066450A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of knowledge tracing, and in particular to a knowledge tracing method based on dual path matrix decomposition. Background Art
[0002] With the development and progress of information technology, traditional learning methods are undergoing profound changes. The most notable sign of this change is the rise of online tutoring systems. Online tutoring systems can accumulate rich learning behavior information by collecting massive amounts of user learning data, which can provide solid data support for knowledge tracking technology. Based on these data, knowledge tracking technology can accurately analyze students' learning progress, knowledge mastery, learning difficulties and other learning situations. The task of knowledge tracking is to track students' mastery of knowledge concepts based on their historical answer sequences and the interaction records generated in past learning activities, so as to predict students' performance in future answering questions.
[0003] The current knowledge tracking models can be mainly divided into two types: classic knowledge tracking models and knowledge tracking models based on deep learning. The existing knowledge tracking models either only focus on the impact of characteristics such as exercise difficulty on future answering performance, or only build characteristics such as learners' knowledge level to enhance the prediction effect. Ignoring one of them will easily lead to a decrease in the accuracy of the prediction results. At the same time, although other learning behaviors of students cannot be directly encoded into the answer sequence, when analyzing the learning situation of the student group, it can be found that behavioral factors such as learners' learning habits and styles may have a certain degree of impact on answering questions.
[0004] Therefore, when conducting attribution analysis on the knowledge and abilities of a learning group, it is necessary to collect as comprehensive information as possible on learners’ learning behaviors on online education platforms and to deeply explore the representation of students’ internal cognitive structures. Summary of the invention
[0005] The embodiment of the present application provides a knowledge tracking method based on dual path matrix decomposition to address the defects of the above-mentioned related technologies. The embodiment of the present application aims to collect students' answer information and as comprehensive behavioral information as possible on the learning platform, construct the parameters about exercises and learners that the model needs to learn, and conduct a more in-depth attribution analysis of students' knowledge mastery. The technical solution is as follows:
[0006] In a first aspect, an embodiment of the present application provides a knowledge tracking method based on dual path matrix decomposition, comprising:
[0007] Obtain the learning record data of each student on the online learning platform;
[0008] Inputting the learning record data of each student into the trained knowledge tracking model, so that the trained knowledge tracking model performs feature extraction based on the learning record data and outputs the learning behavior data of each student; wherein the learning behavior data includes the student feature data and exercise feature data of each student;
[0009] Inputting the learning behavior data and the learning record data of each student into the clustering model, so that the clustering model performs cluster analysis on the learning behavior data and learning data of all students, and outputting the knowledge tracking result label corresponding to each cluster center according to the clustering result;
[0010] The learning record data of each student corresponding to the knowledge tracking result label is analyzed respectively, and the learning data analysis result corresponding to the knowledge tracking result label is output, and corresponding learning guidance information is generated based on the learning data analysis result and fed back to the corresponding students respectively.
[0011] In an optional solution of the first aspect, after obtaining the learning record data of each student on the online learning platform, the method further includes:
[0012] Based on the learning data of each student, obtain the answer record data and the learning log record data of each student, process the answer record data and the learning log record data to obtain the answer sequence of each student, and output the answer data of each student;
[0013] A corresponding knowledge point set and an exercise set are extracted based on the answer record data and the learning log record data, and an exercise-knowledge point association matrix is output based on the association relationship between each exercise and at least one knowledge point.
[0014] In an optional solution of the first aspect, the association relationship between each exercise and at least one knowledge point is determined based on the following steps, including:
[0015] Extracting entities of preset data types from the answer record data and the learning log record data based on the knowledge graph;
[0016] Taking exercises as the main entity and knowledge points as the secondary entities, determine the secondary entities that have an association relationship with each of the main entities, add the association relationship between each main entity and secondary entity that have an association relationship, and obtain multiple triples <exercise, relationship, knowledge point> constructed based on exercises and knowledge points.
[0017] In an optional solution of the first aspect, inputting the learning record data of each student into a trained knowledge tracking model so that the trained knowledge tracking model performs feature extraction based on the learning record data includes:
[0018] Inputting the answer data and the exercise-knowledge point association matrix corresponding to each student into the trained knowledge tracking model, performing feature extraction based on the answer data and the exercise-knowledge point association matrix through the trained knowledge tracking model, and extracting the student feature data and the exercise feature data of each student;
[0019] The student characteristic data include proficiency of knowledge points, mastery of knowledge points, probability of guessing answers, probability of making mistakes in answering questions, and the exercise characteristic data include difficulty of exercises and discrimination of exercises.
[0020] In an optional solution of the first aspect, the knowledge point proficiency is calculated based on the following steps, including:
[0021] Based on the answer data corresponding to each student and the exercise-knowledge point association matrix, the information space R∈{-1,0,1} of all exercises answered by all students at all times is obtained. n×T×m ;
[0022] The answer result vector R itj Multiplying the exercise-knowledge point association matrix to calculate the proficiency of each student's corresponding knowledge point;
[0023] Among them, R itj Indicates students i At time t, for problem q j The answer result vector, R itj =1 means student s i At time t, for problem q j Give the correct answer, R itj =0 means student s i At time t, for problem q j Give the wrong answer, R itj =0 means student s i At time t, for problem q j No answer was given; the total number of students is n, and the total number of exercises is m.
[0024] In an optional solution of the first aspect, the knowledge point mastery degree is calculated based on the following steps, including:
[0025] Get students i The mastery of the knowledge point at time t, obtain the student s i The proficiency of all knowledge points at time t+1;
[0026] Based on students iThe mastery of knowledge points at time t and the proficiency of all knowledge points at time t+1 are calculated for student s i The change in the degree of mastery of knowledge points during the learning process from time t to time t+1;
[0027] Calculate the students' mastery of knowledge points at time t+1;
[0028] The initial knowledge mastery level of the student at the initial moment is determined based on the historical answer data of the student.
[0029] In an optional solution of the first aspect, the exercise feature data is calculated based on the following steps, including:
[0030] Obtaining the answer sequence of each student to each exercise, calculating the number of students who give correct answers to each exercise, and determining the difficulty of the exercise based on the number of students who give correct answers;
[0031] The answer results of multiple students to the same exercise are obtained, the distribution of the answer results and the knowledge points involved in the same exercise are obtained, and the discrimination of the exercise is determined based on the distribution of the answer results, the search for the knowledge points involved in the same exercise and the difficulty of the same exercise.
[0032] In a second aspect, an embodiment of the present application further provides a knowledge tracking device based on dual path matrix decomposition, comprising:
[0033] The first data collection unit is used to obtain the learning record data of each student on the online learning platform;
[0034] A second data collection unit is used to input the learning record data of each student into the trained knowledge tracking model, so that the trained knowledge tracking model performs feature extraction based on the learning record data and outputs the learning behavior data of each student; wherein the learning behavior data includes the student feature data and exercise feature data of each student;
[0035] A data analysis unit, used for inputting the learning behavior data and the learning record data of each student into a clustering model, so that the clustering model performs cluster analysis on the learning behavior data and learning data of all students, and outputs a knowledge tracking result label corresponding to each cluster center according to the clustering result;
[0036] The data analysis unit is also used to analyze the learning record data of each student corresponding to the knowledge tracking result label, output the learning data analysis results corresponding to the knowledge tracking result label, and generate corresponding learning guidance information based on the learning data analysis results and feed it back to the corresponding students.
[0037] In a third aspect, an embodiment of the present application further provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method provided in the first aspect of the embodiment of the present application or any one of the implementation methods of the first aspect is implemented.
[0038] In a fourth aspect, the present application also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method provided by the first aspect of the embodiment of the present application or any one of the implementations of the first aspect.
[0039] The beneficial effects brought about by the technical solutions provided by some embodiments of the present application include at least:
[0040] The embodiment of the present application provides a knowledge tracking method based on dual path matrix decomposition, which further understands the learner's knowledge proficiency and knowledge mastery, as well as the role of factors such as exercise difficulty and discrimination in students' future performance in answering questions by building a knowledge tracking model KTM to learn about learners and exercises. By integrating student behavior data with learner and exercise characteristics, it is possible to more comprehensively represent the learner's learning situation, so that the student group can be classified more accurately, and the learner's learning situation can be further divided, and the learner's answer situation and learning behavior information and learning style, learning habits and other aspects can be matched, so that teachers can more clearly understand the learning status of each student, and adjust the teaching plan according to the student's actual learning status, and can also prompt each learner to understand the shortcomings of their own learning, so that learners can better plan their learning plans according to their actual learning situation. BRIEF DESCRIPTION OF THE DRAWINGS
[0041] In order to more clearly illustrate the technical solutions in the present application or related technologies, the drawings required for use in the embodiments or related technical descriptions are briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0042] Figure 1 It is a flowchart of a knowledge tracking method based on dual path matrix decomposition according to an embodiment of the present application;
[0043] Figure 2 It is a structural schematic diagram of a knowledge tracking device based on dual path matrix decomposition provided in an embodiment of the present application;
[0044] Figure 3 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0045] In order to make the purpose, technical solutions and advantages of this application clearer, the technical solutions in this application will be clearly and completely described below in conjunction with the drawings in this application. Obviously, the described embodiments are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0046] The terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device including a series of steps or modules is not limited to the listed steps or modules, but may optionally include steps or modules that are not listed, or may optionally include other steps or modules that are inherent to these processes, methods, products or devices.
[0047] It should be noted that the terms "first\second" involved in the present application are only used to distinguish similar objects, and do not represent a specific order for the objects. It is understandable that "first\second" can be interchanged with a specific order or sequence where permitted. It should be understood that the objects distinguished by "first\second" can be interchanged where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those described or illustrated herein.
[0048] The present application is described in detail below with reference to specific embodiments.
[0049] Next, combine Figure 1 , taking the calculation method of the usage time of the terminal executing the application as an example, the knowledge tracking method based on dual path matrix decomposition provided by the embodiment of the present application is introduced. Figure 1 , Figure 1 A flow chart of a knowledge tracking method based on dual path matrix decomposition provided by an embodiment of the present application is shown, and the method comprises the following steps:
[0050] S101, obtain the learning data of each student on the online learning platform, and construct the exercise-knowledge point association matrix M respectively k , get the student's answer data D Resp .
[0051] S102, the student's answer data D Resp The exercise-knowledge point association matrix is input into the trained knowledge tracking model, and the learning behavior data of each student is obtained through the trained knowledge tracking model.
[0052] S103, based on the students' learning record data and the characteristics of the students and exercises output by the model, cluster analysis is performed on all students, and a knowledge tracking result label corresponding to each student is output.
[0053] S104, analyzing the student learning records corresponding to various knowledge tracking result labels, determining the influence weights of various indicators on the learning results, and generating corresponding learning guidance results based on the weights.
[0054] Specifically, in S101, the learning data of each student can be obtained through the online learning platform, where the learning data includes but is not limited to the student's reading records of learning materials, the viewing time and viewing progress of learning videos, the number and content of interactions with teachers in online classes, the time for completing exercises and answering records, etc., where the online learning platform can be any network platform for providing content such as exercise questions, video learning materials or text learning materials, so that students can study on the relevant platform and complete the given teaching content, and the embodiments of the present application are not limited to this.
[0055] Specifically, the learning data may include each student's answer record data and learning log record data, wherein the answer record data may include the student's answer content and answer results for each exercise, and the learning log record is used to record the student's learning process data, including but not limited to the learning time of various learning materials, the learning progress of the class, the order of answering questions, the time of answering questions, etc. For example, the above-mentioned learning platform can record each student's mouse click operations in the process of answering specific exercises, the learning time of learning materials (such as text reading time) and other information to record the student's learning behavior, and the embodiments of the present application are not limited to this.
[0056] In some embodiments, after S101, the answer record data R of each student can be obtained based on the learning data of each student. Resp and learning log record data R Log , based on the answer record data R Resp And learning log record data R Log After data preprocessing, obtain the corresponding student's answer sequence:
[0057] X i ={(q1,r1),(q2,r2),(q3,r3),…,(q t ,r t ),…,(q T ,r T )}
[0058] Among them, q t For students i The exercise done at time t, r t For studentsi At time t, exercise q t The answer, r t =1 means the student answered correctly, r t =0 means the student’s answer is wrong.
[0059] For example, as shown in Table 1 below, student s i The answer sequence can be expressed as:
[0060] Table 1 Question answering sequence
[0061]
[0062] In some embodiments, considering that each exercise may correspond to multiple knowledge points, it is difficult to represent the relationship between each knowledge point and exercise through the knowledge graph through the knowledge tracking model. Taking the exercise as the main entity and the knowledge point as the secondary entity, find the secondary entity associated with each main entity and construct <exerciseq i , including knowledge point k i > represents the correspondence between exercises and knowledge points.
[0063] Specifically, the knowledge graph method of exercise-knowledge point is used to process all knowledge points related to all exercises answered by students, and output the knowledge point set C, where |C| = k, c k represents the kth knowledge point in the knowledge point set C; the set Q of all exercises can be expressed as Q = {q1, q2, ..., q j ,…,q M}, so that the corresponding knowledge matrix M can be constructed based on the relationship between each exercise and knowledge point k , through the exercise-knowledge point association matrix, that is, the knowledge matrix M k It can represent the relevance of all knowledge points and exercises involved in all exercises that learners need to answer, including:
[0064]
[0065] in, Each element in the matrix Then it means that exercise q j With knowledge point c k relationship, Representation Exercise q j Involved knowledge point c k ,and Then it means that exercise q j Knowledge point c not involved k .
[0066] It should be noted that one exercise may involve multiple knowledge points, so the knowledge matrix Mk A row in may have more than one non-zero element.
[0067] Further, in S102, the student's answer data D Resp And exercises - knowledge point association matrix M k The data is input into the trained knowledge tracking model KTM, and the learning behavior data of each student is extracted through the trained knowledge tracking model.
[0068] It should be noted that when the students' answer data D Resp And exercises - knowledge point association matrix M k Before inputting into the trained knowledge tracking model KTM, the answer data D Resp And exercises - knowledge point association matrix M k Preprocess the data to meet the format requirements of the input data of the trained knowledge tracking model.
[0069] Specifically, the student's learning behavior data may include student characteristic data and exercise characteristic data, wherein the student characteristic data includes the proficiency of knowledge points, the mastery of knowledge points, the probability of guessing in answering questions, and the probability of making mistakes in answering questions, and the exercise characteristic data includes the difficulty of exercises and the discrimination of exercises.
[0070] Among them, the knowledge point proficiency is used to represent the student's proficiency in each knowledge point. It is mainly obtained based on the analysis of the results of students answering exercises with related knowledge points. If the correct answer ratio is high, the proficiency of the knowledge point is higher. The specific calculation process includes:
[0071] First, obtain the information space R∈{-1,0,1} of all exercises answered by all students at all times n×T×m , R itj Indicates students i At time t, for problem q j The answer result vector, R itj =1 means student s i At time t, for problem q j Give the correct answer, R itj =0 means student s i At time t, for problem q j Give the wrong answer, R itj =0 means student s i At time t, for problem q j No answer was given; the total number of students is n, and the total number of exercises is m.
[0072] The answer result vector R itj With the knowledge matrix M kMultiply them to calculate the students' answers to exercises with related knowledge points, and calculate the knowledge point proficiency M f .
[0073] Furthermore, we can calculate the proficiency of each student's knowledge points according to their f Calculate each student's mastery of all knowledge points M c , where M c The value of is between 0 and 1. The higher the value, the better the students' mastery of all knowledge points. The specific steps include:
[0074] Get students i The degree of mastery of knowledge points at time t Get students i The proficiency of all knowledge points at time t+1
[0075] Based on students i The degree of mastery of knowledge points at time t And the knowledge point proficiency calculation student s for all knowledge points at time t+1 i The change in the degree of mastery of knowledge points during the learning process from time t to time t+1 Application formula:
[0076]
[0077] in, The matrix represents the student s i The sequence of answer results for each knowledge point, The matrix represents each exercise q it The sequence of knowledge points covered;
[0078] Understandably, the change in the degree of mastery of knowledge points is used to characterize the student's i The answers to the exercises during the learning process from time t to time t+1 can indicate that the student s i Changes in the degree of mastery of relevant knowledge points.
[0079] Specifically, the formula can be used to calculate the student's mastery of knowledge points at time t+1: Application formula:
[0080]
[0081] Understandably, Indicates students i The initial knowledge mastery level can be determined based on the student's historical answer data, or a numerical value of the initial knowledge mastery level can be preset, which is not limited in the embodiments of the present application.
[0082] In some embodiments, even if a student answers a question incorrectly during the test, it should be considered that the student's mastery of the knowledge point has changed during the test. Need to have non-negative restrictions, apply the formula:
[0083]
[0084] Among them, ReLU(·) represents a non-negative activation function, which ensures that the obtained change in the degree of mastery of knowledge points can better reflect the changes in the degree of mastery of knowledge points in the learning process.
[0085] Furthermore, we can calculate the number of students who answer each question based on their answers to each question. it The answer result determines the corresponding exercise q it The difficulty and discrimination of exercises; among them, the difficulty of exercises can be calculated based on the number of students who give correct answers to the same exercise, and can be assigned based on the number of students who give correct answers. The more correct answers there are, the lower the difficulty is, and vice versa. The discrimination of exercises can be calculated based on the difficulty of exercises, the distribution of scores of multiple students for the same exercise, and the knowledge points involved in the exercises. If most students' scores are close to the same, it indicates that the discrimination is poor. If the scores of all students are normally distributed, it indicates that the discrimination is good. The discrimination is used to characterize the ability of exercises to test students' mastery of knowledge points and to distinguish between knowledge points that are well mastered and those that are not.
[0086] Specifically, we can get student s i Exercise q answered at time t it All knowledge points involved are calculated using the formula:
[0087]
[0088] Among them, OneHot(·) represents one-hot encoding, Exercise q it After single heat treatment, Trans(·) means transposing the matrix. Representative Exercise q it The relevance to the corresponding knowledge point.
[0089] Further, for Exercise q it , Difficulty and discrimination It can be calculated using the following formula:
[0090]
[0091] in, and are all trainable parameters.
[0092] It should be noted that during the training process of the knowledge tracking model, the knowledge tracking model can be used to predict the student's answer at time t+1 and obtain the probability:
[0093]
[0094] in, It means that based on the student's answers in the previous t moments, the student's answer to question q at time t+1 is predicted. t+1 The answer performance indicates a probability,
[0095] It is understandable that the knowledge tracking model needs to combine the learners' proficiency in knowledge points, the degree of mastery of knowledge points, the difficulty of exercises and the discrimination of exercises to predict relevant parameters. Specifically, the parameterized Rasch model can be used to estimate the students' answers, so as to calculate the probability of guessing and the probability of error in answering questions during the training process.
[0096] Specifically, the formula is applied:
[0097]
[0098] in, It represents a factor vector that combines students’ knowledge proficiency, knowledge mastery, exercise difficulty, and exercise discrimination, which can be used to characterize students’ i Answer exercise q at time t it The complex interaction process between vectors is represented by °, which means the multiplication of elements between vectors.
[0099] Further, based on the above The parametric Rasch model can be used to estimate the students' initial performance:
[0100]
[0101] Among them, P init (s i ,q it ) indicates student s i Answer the Exercise Q it γ is a constant related to a given data set, which can usually be preset to 0 or 0.25. d is a constant in the Rasch model, which can be d = 1.702.
[0102] Furthermore, in the process of students answering questions, it is usually necessary to consider guesswork and error factors to more accurately predict the probability of students answering questions. The formula is applied:
[0103]
[0104] Among them, P(s i ,q it ) indicates student s i Answer the Exercise Q it The final answer probability is represents the overall error probability of students answering all exercises, For students i The overall guessing probability of answering the question, Answer questions for multiple students it The overall guessing probability, error probability and guessing probability can be calculated in the process of the model predicting the students' answering probability.
[0105] Specifically, the above probability is According to the characteristics of the binary classification task of the knowledge tracking model KTM, the loss function is designed as follows:
[0106]
[0107] Among them, r t represents the true answer value of the student at time t, and This means that the knowledge tracking model outputs the student's understanding of exercise q t The goal of the knowledge tracking model is to minimize the difference between the actual answer and the predicted answer, so as to more accurately predict the student's performance in future questions and calculate the probability of error. and guess probability
[0108] Therefore, the above data obtained through the knowledge tracking model include: knowledge point proficiency M f 、Knowledge mastery level M c , Difficulty q diff , discrimination q disc , probability of error and guess probability Combined with the learning log record data R obtained in S101 Log After data processing, the output is the learning behavior data D corresponding to each student. Act .
[0109] Execute S103, based on the knowledge point proficiency M f 、Knowledge mastery level M c , Difficulty q diff , discrimination q discThe guessing probability g and the error probability s are processed to obtain the feature vector of each student, which is input into the clustering model. Based on the clustering result of the clustering model, the knowledge tracking result label corresponding to each student is output, which specifically includes:
[0110] The encoder part of each layer learns the parameters W1 and b1 to obtain the output parameters h. The decoder part of each layer learns the parameters W2 and b2 to obtain the output parameters y. Its goal is to minimize the variance
[0111] The parameters h of the previous layer are used as the input of the next layer, and the ReLU activation function is used in all encoder-decoder pairs (except the parameters g2 of the first pair and g1 of the last pair). The parameters of the model are θ = {W1, b1, W2, b2}.
[0112] In the parameter optimization phase, the soft assignment is iteratively calculated and the KL divergence is minimized, and the auxiliary target distribution is used to learn from the current high-confidence assignment to update the depth map f θ , and refine the cluster centroids.
[0113] Furthermore, the soft assignment of the embedded features to the cluster centroids is calculated, including:
[0114] Preset k cluster centers, the distance between the student's embedded features and the centroid satisfies the t distribution, and the calculation formula for the soft assignment result is as follows:
[0115]
[0116] Furthermore, the auxiliary distribution calculation of the target includes:
[0117] The soft allocation q i Raised to a quadratic power, and then normalized by the frequency of each cluster to calculate the auxiliary distribution p i :
[0118]
[0119] Specifically, the objective function of the deep clustering model DEC is defined as the soft assignment q i and auxiliary distribution p i KL divergence of:
[0120]
[0121] Furthermore, the clustering model DEC jointly optimizes the cluster centers {u j} and the encoder part parameters θ,L with respect to each data point z i and each cluster centroid u j The gradient of the feature space embedding is calculated as follows:
[0122]
[0123]
[0124] Update by back-propagation Thereby achieving the purpose of updating some encoder parameters θ.
[0125] Among them, for the clustering algorithm, the k cluster centroids preset by the clustering model determine the number of cluster centers, and the value of k can be determined by the kL line chart, elbow principle and other indicators.
[0126] Therefore, we can use the learning behavior data of different students D Act Clustering obtains multiple clustering results with the same knowledge tracking result label, and determines the knowledge tracking result label corresponding to each student according to the learning behavior data of each student.
[0127] Specifically, the knowledge tracking result label can be as shown in Table 2 below. The corresponding label is determined according to the specific error probability, guess probability and the learning data of each student obtained from the online learning platform. The learning data may include knowledge points c k The viewing time of related videos, k Text reading time, c k The amount of exercises, c k The degree of knowledge mastery (proficiency in knowledge points and mastery of knowledge points) and the accuracy of answering questions, etc.
[0128] Specifically, the actual error probability, guessing probability, viewing time, text reading time, exercise practice amount, knowledge mastery (knowledge point proficiency and knowledge point mastery) and answer accuracy rate can be compared with the average values of all students respectively. The corresponding threshold can also be preset to determine the difference between the actual value and the threshold, and the corresponding knowledge tracking result label can be determined based on the comparison result and / or the difference with the threshold.
[0129] Table 2 Knowledge tracking result labels
[0130]
[0131]
[0132] The avg superscript indicates the average level of the corresponding data. represents the average accuracy rate, The average time a video is watched. Indicates the average time of text reading, The average value of knowledge mastery, Indicates the average amount of practice for an exercise, “-” indicates that the item is not compared, ~ indicates that it is lower than the corresponding value and the difference with the corresponding value is small, ≈ indicates that the absolute value of the difference with the corresponding value is small, ”>>” indicates that it is greater than the corresponding value and the difference with the corresponding value is large, and “<<” indicates that it is less than the corresponding value and the difference with the corresponding value is large.
[0133] Therefore, the tracking results corresponding to each student can be determined separately, so as to judge the corresponding student's learning results and analyze the specific learning situation. For example, in the first type of tracking results, it can be obviously observed that the students' time investment in learning materials is obviously insufficient, and the amount of exercise practice is also obviously insufficient; for the second type of tracking results, it can be obviously observed that there is a certain lack of time investment in learning materials by students, the amount of exercise practice is not sufficient, the degree of knowledge mastery is slightly lower than the average, and the exercise results are lower than the average; for the third type of tracking results, the error rate of students when doing questions is higher than the average, the time investment in learning materials is obviously higher than the average, the amount of exercise practice is slightly lower than the average, the degree of knowledge mastery is higher than or equal to the average, and the exercise accuracy rate is slightly lower than the average; the fourth and fifth types of tracking results will not be repeated here.
[0134] Furthermore, in step S104, based on the obtained knowledge tracking result labels, learning data analysis results can be output for corresponding students based on the students' learning behavior data, thereby providing learning guidance suggestions for the students.
[0135] Specifically, based on the knowledge tracking result label determined in S103, the common problems of a class of students can be analyzed and determined, so as to provide learning guidance for a class of students, which can not only reflect the common problems of most students, but also improve the efficiency of generating learning guidance.
[0136] Specifically, the learning data analysis results can be in text form. The knowledge tracking result labels output by the knowledge tracking model can be combined with the large language model to construct a prompt template, and the large language model can be used to generate corresponding analysis results and guidance opinions.
[0137] For example, for the first type of tracking results, feedback can be provided such as increasing study time, increasing the amount of exercises practiced, etc. to strengthen learning behavior; for the third type of tracking results, feedback can be provided such as increasing the amount of exercises practiced, and adjusting the knowledge points corresponding to the exercises according to the amount of exercises practiced, etc.
[0138] Specifically, the learning data analysis results can be given to each student based on the student's learning behavior data, as shown in Table 2 below:
[0139] Table 3 Student Group G i Learning data analysis
[0140]
[0141] Specifically, the above-mentioned data analysis projects can be used to determine the impact of students' learning behavior data on learning outcomes, and then weighted calculations can be performed based on the various data in the students' learning behavior data to generate targeted learning guidance results, which is more conducive to pointing out students' deficiencies in specific projects and promoting students' learning.
[0142] Specifically, based on the learning data analysis results of each student, the learning records of students who appear to have better learning effects from a data perspective can be obtained, such as learning time, total exercise volume, amount of exercises for each type of knowledge point, etc. The learning progress of other students can be planned based on the learning records of students with better learning effects. The embodiments of the present application are not limited to this.
[0143] Optionally, the online learning platform can provide a visual interface to facilitate teachers and students to understand the overall learning status.
[0144] The following are device embodiments of the present application, which can be used to execute the method embodiments of the present application. For details not disclosed in the device embodiments of the present application, please refer to the method embodiments of the present application.
[0145] See next Figure 2 , is a schematic diagram of the structure of a knowledge tracking device based on dual path matrix decomposition provided by an exemplary embodiment of the present application. The device can be implemented as all or part of a terminal through software, hardware, or a combination of both, and can also be integrated on a server as an independent module. The knowledge tracking device based on dual path matrix decomposition in the embodiment of the present application can be applied to a terminal or a cloud. The device 20 includes a first data acquisition unit 201, a second data acquisition unit 202, and a data analysis unit 203, wherein:
[0146] The first data collection unit 201 is used to obtain the learning record data of each student on the online learning platform;
[0147] The second data collection unit 202 is used to input the learning record data of each student into the trained knowledge tracking model, so that the trained knowledge tracking model performs feature extraction based on the learning record data and outputs the learning behavior data of each student; wherein the learning behavior data includes the student feature data and exercise feature data of each student;
[0148] The data analysis unit 203 is used to input the learning behavior data and the learning record data of each student into the clustering model, so that the clustering model performs cluster analysis on the learning behavior data and learning data of all students, and outputs the knowledge tracking result label corresponding to each cluster center according to the clustering result;
[0149] The data analysis unit 203 is also used to analyze the learning record data of each student corresponding to the knowledge tracking result label, output the learning data analysis results corresponding to the knowledge tracking result label, and generate corresponding learning guidance information based on the learning data analysis results and feed it back to the corresponding students.
[0150] It should be noted that when the device 20 provided in the above embodiment executes the knowledge tracking method based on dual path matrix decomposition, only the division of the above functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the device provided in the above embodiment and the knowledge tracking method embodiment based on dual path matrix decomposition belong to the same concept. The implementation process is detailed in the method embodiment, which will not be repeated here.
[0151] An embodiment of the present application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of any of the above-mentioned method embodiments when executing the program.
[0152] See also Figure 3 , which is a structural block diagram of an electronic device provided in an embodiment of the present application.
[0153] like Figure 3 As shown, the electronic device 300 includes: a processor 301 and a memory 302 .
[0154] In the embodiment of the present application, the processor 301 is the control center of the computer system, which can be a processor of a physical machine or a processor of a virtual machine. The processor 301 may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor 301 may be implemented in at least one hardware form of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array).
[0155] The processor 301 may also include a main processor and a coprocessor. The main processor is a processor for processing data in the awake state, also called a CPU (Central Processing Unit); the coprocessor is a low-power processor for processing data in the standby state.
[0156] The memory 302 may include one or more computer-readable storage media, which may be non-transitory. The memory 302 may also include a high-speed random access memory, and a non-volatile memory, such as one or more disk storage devices, flash memory storage devices. In some embodiments of the present application, the non-transitory computer-readable storage medium in the memory 302 is used to store at least one instruction, which is used to be executed by the processor 301 to implement the method in the embodiment of the present application.
[0157] In some embodiments, the electronic device 300 further includes: a peripheral device interface 303 and at least one peripheral device 304. The processor 301, the memory 302 and the peripheral device interface 303 can be connected via a bus or a signal line. Each peripheral device 304 can be connected to the peripheral device interface 303 via a bus, a signal line or a circuit board. Specifically, the peripheral device 304 includes: a display screen, a camera and an audio circuit. The peripheral device interface 303 can be used to connect at least one peripheral device related to I / O (Input / Output) to the processor 301 and the memory 302.
[0158] In some embodiments of the present application, the processor 301, the memory 302, and the peripheral device interface 303 are integrated on the same chip or circuit board; in some other embodiments of the present application, any one or two of the processor 301, the memory 302, and the peripheral device interface 303 can be implemented on a separate chip or circuit board. This embodiment of the present application does not specifically limit this.
[0159] The electronic device structure block diagram shown in the embodiment of the present application does not constitute a limitation on the electronic device 300. The electronic device 300 may include more or fewer components than shown in the figure, or combine certain components, or adopt a different component arrangement.
[0160] The embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps of the method of any of the above embodiments are implemented. The computer-readable storage medium may include, but is not limited to, any type of disk, including a floppy disk, an optical disk, a DVD, a CD-ROM, a micro drive, and a magneto-optical disk, a ROM, a RAM, an EPROM, an EEPROM, a DRAM, a VRAM, a flash memory device, a magnetic card or an optical card, a nanosystem (including a molecular memory IC), or any type of medium or device suitable for storing instructions and / or data.
[0161] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, by hardware. Based on this understanding, the above technical solution can be essentially or in other words, the part that contributes to the relevant technology can be embodied in the form of a software product, which can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiment.
[0162] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.
Claims
1. A knowledge tracking method based on dual path matrix decomposition, characterized in that: include: Obtain the learning record data of each student on the online learning platform; Inputting the learning record data of each student into the trained knowledge tracking model, so that the trained knowledge tracking model performs feature extraction based on the learning record data and outputs the learning behavior data of each student; wherein the learning behavior data includes the student feature data and exercise feature data of each student; Inputting the learning behavior data and the learning record data of each student into the clustering model, so that the clustering model performs cluster analysis on the learning behavior data and learning data of all students, and outputting the knowledge tracking result label corresponding to each cluster center according to the clustering result; The learning record data of each student corresponding to the knowledge tracking result label is analyzed respectively, and the learning data analysis result corresponding to the knowledge tracking result label is output, and corresponding learning guidance information is generated based on the learning data analysis result and fed back to the corresponding students respectively.
2. A knowledge tracking method based on dual path matrix decomposition according to claim 1, characterized in that: After obtaining the learning record data of each student on the online learning platform, the method further includes: Based on the learning data of each student, obtain the answer record data and the learning log record data of each student, process the answer record data and the learning log record data to obtain the answer sequence of each student, and output the answer data of each student; A corresponding knowledge point set and an exercise set are extracted based on the answer record data and the learning log record data, and an exercise-knowledge point association matrix is output based on the association relationship between each exercise and at least one knowledge point.
3. A knowledge tracking method based on dual path matrix decomposition according to claim 2, characterized in that: The association relationship between each exercise and at least one knowledge point is determined based on the following steps, including: Extracting entities of preset data types from the answer record data and the learning log record data based on the knowledge graph; Taking exercises as the main entity and knowledge points as the secondary entities, determine the secondary entities that have an association relationship with each of the main entities, add the association relationship between each main entity and secondary entity that have an association relationship, and obtain multiple triples <exercise, relationship, knowledge point> constructed based on exercises and knowledge points.
4. The knowledge tracking method based on dual path matrix decomposition according to claim 2 is characterized in that: The step of inputting the learning record data of each student into the trained knowledge tracking model so that the trained knowledge tracking model performs feature extraction based on the learning record data includes: Inputting the answer data and the exercise-knowledge point association matrix corresponding to each student into the trained knowledge tracking model, performing feature extraction based on the answer data and the exercise-knowledge point association matrix through the trained knowledge tracking model, and extracting the student feature data and the exercise feature data of each student; The student characteristic data include proficiency of knowledge points, mastery of knowledge points, probability of guessing answers, probability of making mistakes in answering questions, and the exercise characteristic data include difficulty of exercises and discrimination of exercises.
5. The knowledge tracking method based on dual path matrix decomposition according to claim 4 is characterized in that: The knowledge point proficiency is calculated based on the following steps, including: Based on the answer data corresponding to each student and the exercise-knowledge point association matrix, the information space R∈{-1,0,1} of all exercises answered by all students at all times is obtained. n×T×m ; The answer result vector R itj Multiplying the exercise-knowledge point association matrix to calculate the proficiency of each student's corresponding knowledge point; Among them, R itj Indicates students i At time t, for problem q j The answer result vector, R itj =1 means student s i At time t, for problem q j Give the correct answer, R itj =0 means student s i At time t, for problem q j Give the wrong answer, R itj =0 means student s i At time t, for problem q j No answer was given; the total number of students is n, and the total number of exercises is m.
6. A knowledge tracking method based on dual path matrix decomposition according to claim 5, characterized in that: The degree of mastery of the knowledge points is calculated based on the following steps, including: Get students i The mastery of the knowledge point at time t, obtain the student s i The proficiency of all knowledge points at time t+1; Based on students i The mastery of knowledge points at time t and the proficiency of all knowledge points at time t+1 are calculated for student s i The change in the degree of mastery of knowledge points during the learning process from time t to time t+1; Calculate the students' mastery of knowledge points at time t+1; The initial knowledge mastery level of the student at the initial moment is determined based on the historical answer data of the student.
7. The knowledge tracking method based on dual path matrix decomposition according to claim 5 is characterized in that: The exercise feature data is calculated based on the following steps, including: Obtaining the answer sequence of each student to each exercise, calculating the number of students who give correct answers to each exercise, and determining the difficulty of the exercise based on the number of students who give correct answers; The answer results of multiple students to the same exercise are obtained, the distribution of the answer results and the knowledge points involved in the same exercise are obtained, and the discrimination of the exercise is determined based on the distribution of the answer results, the search for the knowledge points involved in the same exercise and the difficulty of the same exercise.
8. A knowledge tracking device based on dual path matrix decomposition, characterized in that: include: The first data collection unit is used to obtain the learning record data of each student on the online learning platform; A second data collection unit is used to input the learning record data of each student into the trained knowledge tracking model, so that the trained knowledge tracking model performs feature extraction based on the learning record data and outputs the learning behavior data of each student; wherein the learning behavior data includes the student feature data and exercise feature data of each student; A data analysis unit, used for inputting the learning behavior data and the learning record data of each student into a clustering model, so that the clustering model performs cluster analysis on the learning behavior data and learning data of all students, and outputs a knowledge tracking result label corresponding to each cluster center according to the clustering result; The data analysis unit is also used to analyze the learning record data of each student corresponding to the knowledge tracking result label, output the learning data analysis results corresponding to the knowledge tracking result label, and generate corresponding learning guidance information based on the learning data analysis results and feed it back to the corresponding students.
9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps of the method according to any one of claims 1 to 7 are implemented.
10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.
Citation Information
Patent Citations
Pathway recognition algorithm using data integration on genomic models (PARADIGM)
CN108664762A
Knowledge tracking method and system based on dynamic memory key value pair network
CN116502103A
Precision teaching method for deep knowledge tracking
CN117291775A
Occupational development prediction system and method based on big data analysis
CN117371625A
Method for unsupervised identification of single cell morphological map analysis based on deep learning
CN117788369A
Cited By
Knowledge tracking cold start optimization method and system based on large language model
CN119441508A
